LASAR: Latent Adaptive Semantic Aligned Reasoning for Generative Recommendation

2026/05/31 by Yiwen Chen, Fuwei Zhang, Zehao Chen +7 · 1 citation
Computer Science · #cs.IR

paper · pdf

arxiv created 2026/07/30 · arxiv updated 2026/07/31

Abstract

Large Language Models (LLMs) have demonstrated powerful reasoning capabilities through Chain-of-Thought (CoT) in various tasks, yet the inefficiency of token-by-token generation hinders real-world deployment in latency-sensitive recommender systems. Latent reasoning has emerged as an effective paradigm in LLMs, performing multi-step inference in a continuous hidden-state space to achieve stronger reasoning at lower cost. However, this paradigm remains underexplored in mainstream generative recommendation. Achieving this reveals three key challenges: (1) the gap between prior-less Semantic ID (SID) symbols and continuous latent reasoning, as SIDs lack pre-trained semantics, hindering joint optimization; (2) representation drift due to a lack of reasoning chain supervision; and (3) the suboptimality of applying a globally fixed reasoning depth. To address these, we propose LASAR (Latent Adaptive Semantic Aligned Reasoning), an SFT-then-RL framework. First, we bridge this gap via two-stage training: Stage 1 grounds SID semantics before Stage 2 introduces latent reasoning, ensuring efficient convergence. Second, we mitigate representation drift through explicit CoT semantic alignment. Step-wise bidirectional KL divergence constrains the latent reasoning trajectory using hidden-state anchors extracted from CoT text, while a Policy Head predicts per-sample reasoning depth. Third, during the GRPO-based RL phase, terminal-only KL alignment accommodates variable-length reasoning, and REINFORCE optimizes the Policy Head to dynamically allocate steps. This nearly halves the average latent step count while simultaneously improving recommendation quality. Experiments on three real-world datasets show that LASAR achieves the best overall performance across the evaluated settings. It adds limited inference latency and is roughly 20x faster than generating explicit CoT text.

Citations

• Think-at-Hard: Selective Latent Iterations to Improve Reasoning Language Models
• MiniOneRec: An Open-Source Framework for Scaling Generative Recommendation
• Generative Reasoning Recommendation via LLMs
• OnePiece: Bringing Context Engineering and Reasoning to Industrial Cascade Ranking System
• Bridging Search and Recommendation through Latent Cross Reasoning
• Generative Recommendation with Semantic IDs: A Practitioner's Handbook
• Reinforced Latent Reasoning for LLM-based Recommendation
• R²ec: Towards Large Recommender Models with Reasoning
• LARES: Latent Reasoning for Sequential Recommendation
• Qwen3 Technical Report
• Slow Thinking for Sequential Recommendation
• Rec-R1: Bridging Generative Large Language Models and User-Centric Recommendation Systems via Reinforcement Learning
• Think Before Recommend: Unleashing the Latent Reasoning Power for Sequential Recommendation
• CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillation
• Reasoning with Latent Thoughts: On the Power of Looped Transformers
• SoftCoT: Soft Chain-of-Thought for Efficient Reasoning with LLMs
• Latent Iteration as Renormalization: Inference-Time Recurrence in a Depth-Recurrent Transformer Flows Attention Geometry Toward the SYK Conformal Fixed Point
• Token Assorted: Mixing Latent and Text Tokens for Improved Language Model Reasoning
• Reason4Rec: Deliberative User Preference Alignment of Large Language Models for Recommendation
• Efficient Reasoning with Hidden Thinking
• Towards Large Reasoning Models: A Survey of Reinforced Reasoning with Large Language Models
• Compressed Chain of Thought: Efficient Reasoning Through Dense Representations
• Reasoning Beyond Words ? Exploring framework for hidden state reasoning
• Language Models are Hidden Reasoners: Unlocking Latent Reasoning Capabilities via Self-Rewarding
• Leveraging LLM Reasoning Enhances Personalized Recommender Systems
• Distilling System 2 into System 1
• EAGER: Two-Stream Generative Recommender with Behavior-Semantic Collaboration
• From Explicit CoT to Implicit CoT: Learning to Internalize CoT Step by Step
• Let's Think Dot by Dot: Hidden Computation in Transformer Language Models
• BioXP-0.5B: Explainable Medical-AI via RL-GRPO
• E4SRec: An Elegant Effective Efficient Extensible Solution of Large Language Models for Sequential Recommendation
• Adapting Large Language Models by Integrating Collaborative Semantics for Recommendation
• Implicit Chain of Thought Reasoning via Knowledge Distillation
• Think before you speak: Training Language Models With Pause Tokens
• Graph of Thoughts: Solving Elaborate Problems with Large Language Models
• How Can Recommender Systems Benefit from Large Language Models: A Survey
• How Can Recommender Systems Benefit from Large Language Models: A Survey
• Towards Revealing the Mystery behind Chain of Thought: A Theoretical Perspective
• Tree of Thoughts: Deliberate Problem Solving with Large Language Models
• Recommender Systems with Generative Retrieval
• Large Language Models are Zero-Shot Reasoners
• M6-Rec: Generative Pretrained Language Models are Open-Ended Recommender Systems
• Recommendation as Language Processing (RLP): A Unified Pretrain, Personalized Prompt & Predict Paradigm (P5)
• Self-Consistency Improves Chain of Thought Reasoning in Language Models
• BNAI, NO-TOKEN, and MIND-UNITY: Pillars of a Systemic Revolution in Artificial Intelligence
• BERT4Rec: Sequential Recommendation with Bidirectional Encoder Representations from Transformer
• Self-Attentive Sequential Recommendation
• Session-based Recommendations with Recurrent Neural Networks
• The Expressive Power of Transformers with Chain of Thought
• OpenAI o1 System Card

Cited by

• Hierarchical Latent Reasoning for LLM-based Recommendation

Related

• LaRec: Unleashing LLM-based Latent Reasoning for Generative Recommendation
• Reinforced Latent Reasoning for LLM-based Recommendation
• Generative Reasoning Recommendation via LLMs
• Hierarchical Latent Reasoning for LLM-based Recommendation
• WhisperRec: Latent Reasoning for Efficient Foundation Recommendation Models